๐Ÿ”ฅ Anthropic Study: Experts Followed Their AI's Random Picks 86% of the Time #AINews

Super Intelligence Spotlight
8 views โ€ข

๐Ÿ”ฅ ๐—”๐—ป๐˜๐—ต๐—ฟ๐—ผ๐—ฝ๐—ถ๐—ฐ ๐—ฆ๐˜๐˜‚๐—ฑ๐˜†: ๐—˜๐˜…๐—ฝ๐—ฒ๐—ฟ๐˜๐˜€ ๐—™๐—ผ๐—น๐—น๐—ผ๐˜„๐—ฒ๐—ฑ ๐—ง๐—ต๐—ฒ๐—ถ๐—ฟ ๐—”๐—œ'๐˜€ ๐—ฅ๐—ฎ๐—ป๐—ฑ๐—ผ๐—บ ๐—ฃ๐—ถ๐—ฐ๐—ธ๐˜€ ๐Ÿด๐Ÿฒ% ๐—ผ๐—ณ ๐˜๐—ต๐—ฒ ๐—ง๐—ถ๐—บ๐—ฒ Anthropic set up each analyst's agent to favor one option on each decision, which the analysts weren't told about, decided by coin flip. Analysts went with their agent's pick 86% of the time, against about 50% if the agent's input made no difference. Analyses with the agent had serious flaws 6% of the time, against 55% without it. The median time with the agent was 65 minutes, against 90 minutes without it. 77% of analysts used the agents' own arguments, but only 15% credited the agent as the source. Anthropic says the no-agent group was small and less experienced. #AI #Anthropic #Claude โ”€โ”€โ”€ ๐—™๐—ผ๐—น๐—น๐—ผ๐˜„ ๐—ฆ๐—œ ๐—ฆ๐—ฝ๐—ผ๐˜๐—น๐—ถ๐—ด๐—ต๐˜ ๐—ณ๐—ผ๐—ฟ ๐—ฆ๐—œ/๐—”๐—œ ๐—ฎ๐—ป๐—ฑ ๐˜๐—ฒ๐—ฐ๐—ต ๐—ป๐—ฒ๐˜„๐˜€: ๐—ซ ยท ๐—Ÿ๐—ถ๐—ป๐—ธ๐—ฒ๐—ฑ๐—œ๐—ป ยท ๐—ง๐—ฒ๐—น๐—ฒ๐—ด๐—ฟ๐—ฎ๐—บ #Shorts

Keyboard shortcuts

On. Switch them off if they collide with your assistive tools; ? still opens this sheet.

Go to

Press g then the letter.

  • gh Latest
  • gs Sources
  • gm Media
  • gv Videos
  • gp Podcasts
  • gc Calendar
  • gd Decoder
  • gz Dataviz
  • ga Datasets
  • gb Blog
  • gk Markets
  • gj Careers
  • gn Prompt Notebook

On this page

  • / Focus search, where there is one
  • t Back to top
  • ? This list
  • Esc Close